Abstract: Edge detection of an image is a fundamental, yet one of the most important fields in image processing, pattern recognition, and image analysis and computer vision technique. In edge detection, the structural information of an image is extract and reduced to important data to process. There are different methods for detecting the edge, and mainly categories into search based and zero-crossing based. One of the most common methods is ‘Canny edge detector’, and work well under some condition. Like every other method, it had some disadvantages, drawbacks and limitation. Due to this limitation and constant evolution of image processing, a better method is in demand and a number of researches had been done to improve the methods and the performance of this technique. To rectify the problem, and to have a better outcome, a new method of edge detection is needed. To increase the performance of the detector, the removal of noise from an image is the first important thing such that it does not affect the detection by showing false noise. The detector must also have a single response of thin line for each detected edge. In other words it must present clear and corrected image which might further be used for different purpose: medical field for identifying diseases, object identification and computer vision system etc.
Keywords: Edge detection, image processing, pattern recognition, computer vision technique, Canny detector, 2-D Otsu’s method, Threshold.